Best Power Supply Buying Guide (Jun 2020)

[Caring1]

JonnyGuru

The name rings a bell, isn't he some redundant cartoon character?

 

[duynguyenle]

He himself said the Corsair PSUs aren't great.

Corsair VS Series 650 W Review

At just $55, the Corsair VS650 is very affordable and performs well enough to be your next power supply if you are on a tight budget and your system is not all that power hungry. It aced our tough tests, and comes with 80 PLUS and Cybenetics ratings.

www.techpowerup.com

Corsair RM750 750 W Review

The Corsair RM750 only costs slightly less than the RM750x, and the latter comes with the topnotch NR135L fan and Japanese caps. That said, the RM750 has jaw-dropping efficiency at very light and light loads and is compatible with the new Modern Standby mode in Windows 10, meaning it can wake up...

www.techpowerup.com

Corsair AX Series 1000 W Review

At $240, the Corsair AX1000 is certainly not cheap, but it is the best PSU we ever tested. Being built on a Seasonic platform, only top-notch components are used, making this one of the most recommended 1000 W PSUs you can get today, if your system needs much power.

www.techpowerup.com

Reviewers cannot check products for lifespan.

These Corsair RM, VS (and possibly other models) offer "average" performance and die after a year or two.

2 failed on me and I always see posts of other people having similar issues.

It's always a bit funny when I see posts like this "I always see posts of other people having similar issues". Let me introduce you to the concept of selection bias. Simply put, you see people making complaints post because only people having issues would bother making posts about it. People with perfectly functioning PSUs are much less inclined to make posts about the product they bought funtioning as intended, whereas people with failed units are much more likely to make posts about the products they bought which is NOT functioning as intended.

There are no meaningful conclusions you can draw regarding the true overall failure rate of a product from simply observing the number of posts complaining about failures, because of selection bias.

Speaking of, there seems to be some weird bias in the reviews you linked as well, two out of the three reviews are positive (the AX and RM products), and the conclusion on the VS is that is is affordable and is capable of delivering the full rated power. Where in the review does it say the RM unit is average and die after a year or two?

While we're comparing anectdotes, let me offer my own experience. I've owned a couple of Corsair units over the years, a (now 11 year old) first generation HX850 from way back in 2010 which is still powering my brother's rig, a RM750x from circa 2016 or so which is still in my personal rig. A SF600 gold for the livingroom ITX little computer (bought secondhand, 2017 production), a RM850 in a PC I built for a friend. All fully functioning and I have not had to RMA any of them.

Of course, all of that means didly squat if you want to draw conclusions about overall failure rates across all their product lines (for the reasons above)

 

[Regeneration]

RMA numbers are considered "confidential" for hardware manufactures as a protection against lawsuits and class-action lawsuits that can cost them billions. A single "bad" product can cause irreversible damage to the company's reputation.

The best source of such information is feedback on the Internet. You don't have to be an expert/reviewer/analyst to know that stuffed crust is the most loved pizza, Nvidia's Fermi GPUs tend to fail after several years of use, IBM Deskstar 75GXP HDDs aren't reliable and Corsair VS and RM series aren't great PSUs.

 

[Vayra86]

It's always a bit funny when I see posts like this "I always see posts of other people having similar issues". Let me introduce you to the concept of selection bias. Simply put, you see people making complaints post because only people having issues would bother making posts about it. People with perfectly functioning PSUs are much less inclined to make posts about the product they bought funtioning as intended, whereas people with failed units are much more likely to make posts about the products they bought which is NOT functioning as intended.

There are no meaningful conclusions you can draw regarding the true overall failure rate of a product from simply observing the number of posts complaining about failures, because of selection bias.

Speaking of, there seems to be some weird bias in the reviews you linked as well, two out of the three reviews are positive (the AX and RM products), and the conclusion on the VS is that is is affordable and is capable of delivering the full rated power. Where in the review does it say the RM unit is average and die after a year or two?

While we're comparing anectdotes, let me offer my own experience. I've owned a couple of Corsair units over the years, a (now 11 year old) first generation HX850 from way back in 2010 which is still powering my brother's rig, a RM750x from circa 2016 or so which is still in my personal rig. A SF600 gold for the livingroom ITX little computer (bought secondhand, 2017 production), a RM850 in a PC I built for a friend. All fully functioning and I have not had to RMA any of them.

Of course, all of that means didly squat if you want to draw conclusions about overall failure rates across all their product lines (for the reasons above)

Its a matter of experience, fact and source checking, as it is with everything you read and hear. Anecdotal sources are sources just like any other, and full RMA rates are not a conclusive number either. Its also important WHY units fail, for example. And the VS unit is not suitable for every situation, so this can be a reason to avoid it. User experiences bring them to light more easily than a review might do. Many reviewers miss things. Many users are a lot more motivated to find out everything they can about a unit.

I've drawn a wealth of useful and proven conclusions from this forum and single user experiences. The value is immense.

 

[UK-Antec Rep]

So much bias here it is unreal.

 

[Regeneration]

So much bias here it is unreal.

If its going to make you feel any better, I replaced the failing Corsair PSUs with Antec EarthWatts for the 7 year warranty.

 

[duynguyenle]

Its a matter of experience, fact and source checking, as it is with everything you read and hear. Anecdotal sources are sources just like any other, and full RMA rates are not a conclusive number either. Its also important WHY units fail, for example. And the VS unit is not suitable for every situation, so this can be a reason to avoid it. User experiences bring them to light more easily than a review might do. Many reviewers miss things. Many users are a lot more motivated to find out everything they can about a unit.

I've drawn a wealth of useful and proven conclusions from this forum and single user experiences. The value is immense.

For sure, and I have as well (mainly coming from custom MKB boards/forums). However, one must be constantly aware of selection bias and maybe consider assigning a weighted score accordingly. I still maintain there's nothing wrong with RM units based on my own experiences with them, what weighted score should that comment be assigned?

The issue with relying on user reports to try and find out exactly which parts failed when a unit goes bad is that in most cases the average users don't really have the expertise and equipment to disassemble the failed PSU and diagnose what went wrong. Is it a case of a single faulty component or a design issue that affect the entire line of products? In aerospace, incident reports are usually quite detailed, and the failure assessment process is usually specifically geared to root out this type of issue ('freak accident' vs endemic underlying issues with the design). I don't feel like we have this level of standardisation and detail work for analysing failed PSUs.

For the most part, I agree the VS units are rather mediocre and I'd probably look elsewhere.

 

[Regeneration]

Corsair RM series after some time (a year or two) tend to be unstable under heavy load scenarios (e.g. gaming).

You play a game and all of the sudden: hard lockup / shutdown / reboot / black screen. Now that's not pretty sight.

Then you spend hours finding out why. "But that Corsair RM PSU, i bought it just 1-2 years ago."

 

[duynguyenle]

Corsair RM series after some time (a year or two) tend to be unstable under heavy load scenarios (e.g. gaming).

You play a game and all of the sudden: hard lockup / shutdown / reboot / black screen. Now that's not pretty sight.

Then you spend hours finding out why. "But that Corsair RM PSU, i bought it just 1-2 years ago."

That's interesting. I'll keep an eye out for that. Do we know a root cause/hardware combination for this behaviour under load i.e. what hardware combination typically triggers that, what load condition (CPU heavy or GPU heavy). Is it because voltage goes out of spec intermittently or something else (excessive ripple causing instabilities). Does heavy overclock accelerate failure or cause it to happen more often?

I built a PC for a friend with a RM850, currently running a fairly highly overclocked 9700K and 2080S, I'll check if this has manifested.

 

[Regeneration]

That's interesting. I'll keep an eye out for that. Do we know a root cause/hardware combination for this behaviour under load i.e. what hardware combination typically triggers that, what load condition (CPU heavy or GPU heavy). Is it because voltage goes out of spec intermittently or something else (excessive ripple causing instabilities). Does heavy overclock accelerate failure or cause it to happen more often?

I built a PC for a friend with a RM850, currently running a fairly highly overclocked 9700K and 2080S, I'll check if this has manifested.

GPU heavy and the root cause its just degrading over time. On day-1 its 1000% all perfect but after a year not so much.

OCed i7-9700K and 2080 Super deserve high-end PSU.

 

[duynguyenle]

GPU heavy and the root cause its just degrading over time. On day-1 its 1000% all perfect but after a year not so much.

OCed i7-9700K and 2080 Super deserve high-end PSU.

And does this affect old generation RM units only or the refreshed line as well? (I am not sure how many refreshes the RM line has had, the one I bought is from back when they still had the yellow lettering accent colour.

Do we have any data on which component degradation causes this? Might be a good excuse for me to get that hot air station I've been eyeing for a while. If the component is somewhere accessible and not glued down I can probably replace it pretty easily.

 

[ShurikN]

That SF recommendation didn't age well

 

[Chrispy_]

GPU heavy and the root cause its just degrading over time. On day-1 its 1000% all perfect but after a year not so much.

OCed i7-9700K and 2080 Super deserve high-end PSU.

I genuinely don't understand why you're seeing this behaviour from RM supplies. There are three lines at this point - the original RM750 for example, then RM 750i and now the RM750X. Several sites have reviewed different variants and wattages and they switched early from Chicony to CWT as the manufacturer shortly after the RM was launched.

First of all, the Chicony originals were good.
Secondly, the CWT designs for the vanilla RM, i-series, and X-series have also been good. No bad reviews, no bait-and-switch components inside, no corners being cut on component supplier or rating of components chosen from any of those suppliers.

Power supplies aren't magic - they are the sum of their parts, design, and build quality. RM supplies are using tried-and-true CWT designs known to be reliable elsewhere - they are assembled to a decent quality with good parts from reliable higher-tier suppliers like Nippo Chemi-con, Rubycon and Weltrend - exactly the same as other leading PSU manufacturers.

• So the parts are known to be high-quality and used industry-wide in millions of power supplies from around the globe; they're not unique to Corsair or CWT.
• The platform design is conservative and not unique to the RM series or even Corsair. CWT will use that platform across multiple other brands they build. Corsair is just one of CWT's many customers.
• The assembly quality of all reviewed units has been somewhere between "good" and "perfect".
• The warranty is long. If there was a design flaw that caused a high prevalance of regular failures after one year, Corsair would likely change platform, but they haven't.

The only thing that really makes them a "Corsair RM" power supply is that CWT are assembling them to corsair's spec with Corsair RM logos and packaging. There is no reason why Corsair RMs should be any more or less failure prone than the dozens of other models or other brands that also share that platform design from CWT.

....and no, I don't work for Corsair or CWT. I'm not even a particularly big fan or Corsair - I think they're pretty mediocre but they are at least widely available and don't make anything too terrible which makes them an easy, safe recommendation most of the time.

 

[duynguyenle]

I genuinely don't understand why you're seeing this behaviour from RM supplies. There are three lines at this point - the original RM750 for example, then RM 750i and now the RM750X. Several sites have reviewed different variants and wattages and they switched early from Chicony to CWT as the manufacturer shortly after the RM was launched.

First of all, the Chicony originals were good.
Secondly, the CWT designs for the vanilla RM, i-series, and X-series have also been good. No bad reviews, no bait-and-switch components inside, no corners being cut on component supplier or rating of components chosen from any of those suppliers.

Power supplies aren't magic - they are the sum of their parts, design, and build quality. RM supplies are using tried-and-true CWT designs known to be reliable elsewhere - they are assembled to a decent quality with good parts from reliable higher-tier suppliers like Nippo Chemi-con, Rubycon and Weltrend - exactly the same as other leading PSU manufacturers.

• So the parts are known to be high-quality and used industry-wide in millions of power supplies from around the globe; they're not unique to Corsair or CWT.
• The platform design is conservative and not unique to the RM series or even Corsair. CWT will use that platform across multiple other brands they build. Corsair is just one of CWT's many customers.
• The assembly quality of all reviewed units has been somewhere between "good" and "perfect".
• The warranty is long. If there was a design flaw that caused a high prevalance of regular failures after one year, Corsair would likely change platform, but they haven't.

The only thing that really makes them a "Corsair RM" power supply is that CWT are assembling them to corsair's spec with Corsair RM logos and packaging. There is no reason why Corsair RMs should be any more or less failure prone than the dozens of other models or other brands that also share that platform design from CWT.

....and no, I don't work for Corsair or CWT. I'm not even a particularly big fan or Corsair - I think they're pretty mediocre but they are at least widely available and don't make anything too terrible which makes them an easy, safe recommendation most of the time.

This is my experience with the RM line too, but clearly I haven't been reading enough forum threads about them to understand why they're falling.

In terms of power electronics, I think electrolytic caps are the most liable to 'degrade' in the ways Regeneration seems to describe, but most caps in modern PSUs aren't electrolytic (I think it's mainly just the bulk hold-up caps) and even then, as capacitor ages it shouldn't cause the kind of failures like that (and if a cap does blow up, you'll just end up with a PC that doesn't boot instead of one that only acts up under certain workloads). Aside from the fan, a PSU is mainly just solid state electronics, there aren't that many failure modes (assuming the PSU uses a mature topology and good components as you pointed out)

Also I'm interested to know the differences between a heavy CPU workload and GPU work load in terms of load demand and how heavy GPU work is more likely to trigger failures. As far as I know, they all look like 12v loads to the PSU. Is there any differences in the way GPU and CPU pulls power? And how does this contribute to failure?

All of this is rather intriguing, as someone said in their post, end-users would be very interested in finding out exactly what went wrong, unfortunately I'm not getting the full picture here.

 

[Regeneration]

A new thread from earlier today:

my pc keeps restarting while im playing fortnite

So, Im new to this pc but I still know all my specs so bare with me. I started as a slow reoccurring theme, while I would be playing my pc would randomly reboot no blues screen nothing just restarting. I have done tons of research and some said could be faulty psu faulty mother board but as you...

www.techpowerup.com

Anyone wants to make a wild guess?

 

[duynguyenle]

A new thread from earlier today:

my pc keeps restarting while im playing fortnite

So, Im new to this pc but I still know all my specs so bare with me. I started as a slow reoccurring theme, while I would be playing my pc would randomly reboot no blues screen nothing just restarting. I have done tons of research and some said could be faulty psu faulty mother board but as you...

www.techpowerup.com

Anyone wants to make a wild guess?

That doesn't really answer my question though. I would like to know the scenarios this happens. Is it because of voltage regulation? Improperly configured factory fan curve leading to OTP being tripped? Caps going bad?

Without more information, it's not possible for me to try and figure out the root cause (because some of these scenarios have a simple fix, and some do not), this is especially unhelpful to me since none of the RM units that has passed my hands exhibited this behaviour (even the ones that's powering this very rig I'm typing this comment on. The user of this thread you linked has a RMx series PSU, do you think this is the same type of failure that apparently plagues RM units? If yes, then we can probably start ruling out some components being the culprit. I think they're both made by the same ODM (Channel Wells) but are they on the same platform (which revision of the RMx PSU is this? IIRC the RMX line has been refreshed once I think).

Can't really make a wild guess without more information. I'll see if the guy can give more detailed description of his specs

 

[djisas]

I had been looking at corsair psus before buying, I think I dodged a landmine...

 

[cucker tarlson]

when are you gonna review a BeQuiet/FSP PSU ?

 

[juular]

Seasonic for ex renewed it's entire lineup last year...

All they did is replacing 'Focus Plus Gold' and 'PRIME Ultra Gold' with 'Focus GX' and 'PRIME GX' on the label, no significant differences inside or differences in performance (aside of fix for shutting down with AMD Vega which perhaps could've been fixed by just moving some traces around) as far as i'm aware. Well, and added a couple more fanless SKUs to the portfolio while at it.

Corsair RM series after some time (a year or two) tend to be unstable under heavy load scenarios (e.g. gaming).

First, if you're talking about 'Corsair RM', you gotta specify which ones. Old Corsair RM from 2013 ? New Corsair RM from 2019 ? Old Corsair RMx from 2015 ? New Corsair RMx from 2018 ? Perhaps Corsair RMi ?
Second, surely, there are no such threads with any other PSUs involved, only Corsair ones. /s
Third, in all cases i've seen if there are lock-ups or restarts - it's either a completely shitty PSU like some bargain bin shit or things like EVGA N1, or it's not a PSU problem - RAM instability, CPU OC, GPU instability due to age or OC, you name it. Well, sometimes it can be a kinda PSU problem due to corroded/overload cables, kinda - because it's a user to blame here (do not overload cables, check your cables for corrosion or avoid moisture buildup if you live in humid conditions).
Fourth, Corsair PSUs, CWT or not are made by same OEMs as quite a few of other brand lineups, so in fact you're talking not Corsair PSUs but that OEM's PSUs, assuming these problems are indeed present which i highly doubt.
Fifth, selection & confirmation bias exists, you're seeing what you want to see.

 

[cucker tarlson]

rm650x, focus 650 or superflower leadex 3 650 ? which one to get. I require very quiet.

 

[juular]

rm650x, focus 650 or superflower leadex 3 650 ? which one to get. I require very quiet.

If it's a Focus Platinum then really, either of them. Otherwise either of other two, they're pretty much on par both electrical and acoustic performance-wise. Although Super Flower had some problems with early revisions of Leadex III so perhaps Corsair is the winner here if you're not sure if Leadex would be from recent batch.

 

[cucker tarlson]

If it's a Focus Platinum then really, either of them. Otherwise either of other two, they're pretty much on par both electrical and acoustic performance-wise. Although Super Flower had some problems with early revisions of Leadex III so perhaps Corsair is the winner here if you're not sure if Leadex would be from recent batch.

650x is 560 pln, leadex III is 600, focus px is 650.
worth a small premium over rm650x ?

 

[juular]

650x is 560 pln, leadex III is 600, focus px is 650.
worth a small premium over rm650x ?

Poland ? What about be quiet! Straight Power 11 or SilentiumPC Supremo FM2 then ? If not, well, Corsair looks like an obvious choice since it's cheaper.

 

[cucker tarlson]

Straight Power 11 Plat ? I read the build quality is excellent but performance is a notch worse than SS/RMx/SFl

same price as 650x for SP11 650W Plat

 

[juular]

Straight Power 11 Plat ? I read the build quality is excellent but performance is a notch worse than SS/RMx/SFl

Gold one. But Platinum would be really okay if it's cheapest option too.
Also a few other random suggestions : Bitfenix Whisper M (if you can find it in stock), Deepcool DQ-M, Fractal Design ION+.